Human Identification with Electrocardiogram Signals: a Neural Network Approach

This paper presents a neural network developed to identify human subjects using electrocardiogram (ECG) signals collected from an 'in-house' wearable electrocardiogram (ECG) sensor. In this project, noises were first removed from the raw signals with wavelet filters. ECG cycles were then e...

Full description

Saved in:
Bibliographic Details
Published in:AIP conference proceedings 2008-10, Vol.1127, p.19-27
Main Authors: Wan, Yongbo, Yao, Jianchu
Format: Article
Language:English
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper presents a neural network developed to identify human subjects using electrocardiogram (ECG) signals collected from an 'in-house' wearable electrocardiogram (ECG) sensor. In this project, noises were first removed from the raw signals with wavelet filters. ECG cycles were then extracted from the filtered signals and decomposed into wavelet coefficient structures. These coefficient structures were used as input vectors to a 3-layer feedforward neural network that generates the identification results. In the current study, 61 datasets collected from 23 subjects were utilized to train the neural network, which thereafter was tested with 15 new datasets from 15 different subjects. All the 15 subjects in the experiment were successfully identified. The testing results demonstrate that the neural network is effective.
ISSN:0094-243X